Error prediction of VMD-GRU current transformer based on bayesian optimization

Abstract

Abstract With the aim of accurately predicting the measurement error of the current transformer, a Bayesian optimization-based error prediction method combining variational mode decomposition (VMD) and gated recurrent unit (GRU) was proposed. Firstly, the secondary current signal of the CT is decomposed by VMD, and the arrangement entropy of each sequence signal is calculated, so as to determine the best decomposed mode number. Then, the decomposed modal components are utilized as inputs to the recognition model. Finally, a GRU model serves as the prediction model for each component, forecasting individual components, after which the predicted results of these components are reconstructed to yield the ultimate prediction outcome.